Parameter Free Image Artifacts Detection: A Compression Based Approach
|
|
- Samuel Allison
- 6 years ago
- Views:
Transcription
1 Parameter Free Image Artifacts Detection: A Compression Based Approach Avid Roman-Gonzalez a*, Mihai Datcu b a TELECOM ParisTech, Paris, France b German Aerospace Center (DLR), Oberpfaffenhofen ABSTRACT The qualified Earth Observation (EO) images sometimes may present unexpected artifacts. These perturbations and distortions can make more difficult the analysis of images and may decrease the efficiency of interpretation algorithms because the information is distorted. Thus is necessary to implement methods able to detect these artifacts regardless of the model which are formed, i.e. parameter free. In this article, we propose and present a method based on data compression, whether lossy compression or lossless compression for detecting aliasing, strips, saturation, etc. 1. INTRODUCTION The qualified EO images, sometimes may present unexpected artifacts, distortions or artificial structures which can be produced by the instrument itself or due to pre-processing. These perturbations may come from miss adaptation of the sensor resolution for the scene detected, thus aliasing may occur, artifacts due to eventual compression of the EO product, potential saturation of the sensor for particular image geometry, accidental image stripes or bands, etc. These perturbations and distortions can make more difficult the analysis of images and may decrease the efficiency of interpretation algorithms because the information is distorted. The artifacts produce very small variations in gray level. The growing volume of data provided by different sources of images requires the use of tools to perform analysis automatically, such as similarity detection, classification, pattern recognition, etc. Fig 1 shows some examples of artifacts, in (a) and (b) we can see a change of texture; (c) and (d) show the existence of horizontal lines which can be detected as bridges; (e) shows the saturation and (f) shows the blocking. All these processes may be affected if the image has artifacts, these artifacts may interfere with the recognition of a texture or the quantification of features. The nature of some of these artifacts is unknown and varied, these artifacts doesn t have the same models, that is why it is necessary to implement methods able to detect these artifacts regardless of the model which are formed, i.e. parameter free. In the process of elaborating the standard product of a satellite image, there is a process of correction of artifacts as presented by Hyung-Sup in [1] which is an algorithm for the restoration of defectives lines; however some artifacts may be remaining after this process. (a) (b) * roman@telecom-paristech.fr; phone Image and Signal Processing for Remote Sensing XVI, edited by Lorenzo Bruzzone, Proc. of SPIE Vol. 7830, SPIE CCC code: X/10/$18 doi: / Proc. of SPIE Vol
2 (c) SPOT (d) (e) IKONOS (f) MERIS Fig. 1 Some Examples of Artifacts: (a) and (b) Shows a change of texture by aliasing. (c) Shows a sea picture where there a bridge. (d) Shows the existence of horizontal lines and these lines could be detected as bridges by an algorithm for indexing. (e) Shows saturation of the sensor. (f) Shows the blocking in the picture. The work presented in this paper is the continuation of the work by A. Mallet in [2] and [3]. In this article, we propose the use of compression techniques, lossless and lossy compression, such as parameter-free method for the artifacts detection like aliasing, strips, saturation, etc. Using these compression techniques the aim is to evaluate the level of regularity or irregularity that may have an artifact. In this article we propose two methods: The first method uses Lossy Compression to calculate the Rate-Distortion Function. The Rate-Distortion analysis allows us to evaluate how much the data was distorted, we further develop and asses the method in [3] based on the analysis of the lossy compression error for variable compression factor. The error behavior of the sectors with artifacts is different from the sectors that do not contain artifacts. Second method uses the Lossless Compression to calculate Normalized Compression Distance (NCD), the NCD is a method proposed by M- Li in [4] to determine the similarity between two files using a distance measure based on Kolmogorov complexity. The paper is structured as follow. Section II presents an overview on the theory on which these methods are based. Section III shows practical applications in artifacts detection in optical images. Finally section IV reports our conclusions. 2. THEORETICAL BACKGROUND 2.1 Rate Distortion Function The Rate-Distortion (RD) Function is given by the minimum value of mutual information between source and receiver under some distortion restrictions. Proc. of SPIE Vol
3 R ( D) = min Q QD I ( p, Q) Where: I ( p, Q) Is the Mutual Information between p and Q. The RD function shows the compression error given for different compression factor. The RD function depends of the visual complexity, for example plotting Experimental RD Curve, we can do an analysis of the image, for example Fig. 2 shows two images in (a) shows an image that has a lot of detail because it contains hills, houses, trees, cars, people, etc; while in (b) we have a picture without much detail that the background appears to be constant and the color of the dogs too. This analysis is reflected in the experimental RD curve shown in Fig. 3 in which we can see that the curve of the image (a) which is the blue line is above to the curve of the image (b) which is the green. Can be considered that the artifacts increase the complexity in the image, for example we can do the same experiment to see the change in the experimental RD curve of any image and the same image but with synthetic lines added, in Fig. 4 (a) shows the normal Cusco city image and Fig. 4 (b) shows the Cusco city image but with some defectives lines. We can observe in Fig. 5 the experimental RD curves for each image, blue for (a) and green for (b), we can see that the curve of the image with defectives lines is above to the curve of the normal image, this is because the defects cause that the image is more irregular. To calculate RD Function, is used Baseline-JPEG as a lossy compressor. (a) Cusco City (b) Puppies Fig. 2: (a) Cusco City is an image with a lot of detail; it contains hills, houses, trees, cars, people, etc. (b) Puppies is an image with a background almost constant, this image is less complex than image (a) (a) (b) Fig. 3 Experimental Rate-Distortion Curve: blue for Fig. 2 (a) and green for Fig. 2 (b), the experimental RD curve for the complex image (a) is above than the experimental RD curve for image (b). Proc. of SPIE Vol
4 (a) Cusco City (b) Cusco City with Defectives Lines Fig. 4: (a) Cusco City is an image normal. (b) Is the same image (a) but with defectives lines artificially introduced to analyze its influence to the calculation of experimental RD curve (b) (a) Fig. 5 Experimental Rate-Distortion Curve: blue for Fig. 4 (a) and green for Fig. 4 (b), the experimental RD curve of normal image is below than the experimental RD curve of image with defectives lines 2.2 Normalized Compression Distance (NCD) The Kolmogorov Complexity K(x) of an object x is the minimum amount of computational resources (q) needed to represent x. The Kolmogorov Complexity is defined by: K ( x ) = min q q Qx Where: Qx Is the set of instantaneous codes that give as output x. There is a dependency of the size and descriptive language, but it is not very worrying as it is reduced to some constant, i.e., given two languages L1 and L2, and any string of symbols x, K1(x)-K2(x) < k. For move from a description in L1 to another in L2 is a program interpreter of L1 in L2 writing. The interpreter may be more or less long, but it's fixed, so that its size is a constant. Within the Information Theory we can say that the Kolmogorov complexity or algorithmic complexity is the amount of information needed to recover x. It is important to note that K(x) is a non-calculable function. The conditional complexity K(x, y) of x related to y is defined as the length of the shortest program with which we can obtain an output x from y. An important application of this notion is to estimate the shared information between two objects: The Normalized Information Distance (NID) [1]. The NID is proportional to the length of the shortest program that can calculate x given y. The distance calculated from these considerations is then normalized as follows: Proc. of SPIE Vol
5 (1) NID( x, y) { } { } (, ) min ( ), ( ) max K( x), K( y) K x y K x K y = The NID result is a positive value r in the range of: 0 r 1, with r = 0 if the objects are identical and r = 1 is the maximum distance between them. The compressibility is a term that derives from having a small program P that describes a string x. Therefore, if K(x) < x we say that x is compressible. The Normalized Compression Distance NCD has a definition as follows: Since the Kolmogorov Complexity K(x) is a non-computable function, an approximation is defined by Li in [1] considering K(x) as the compressed version of x, and a lower limit of what can be achieved with the compressor C. Previous research [1, 2] use the Kolmogorov Complexity for define the Normalized information Distance NID and from there derive the Normalized Compression Distance. It is to approximate K(x) with C(x) = K(x) + k, the length of the compressed version of x obtained by a lossless compressor C plus an unknown constant k: the presence of k is necessary because it is not possible to estimate how close of K(x) is this approach. To clarify this concept, we take two strings b y p have the same length n, where the first is a random output of a Bernoulli process and the second represents the first n digits of π. The quantity K(p) would be smaller tan K(b) because there is a program of length K(p) << n which output is the number π, while a program that has as output a random sequence of bits would have a close to n length, is so K(p) << K(b). Thus, the equation (1) could be estimated by the Normalized Compression Distance (NCD) as shown in equation (2): (2) NCD( x, y) = { } { } (, ) min ( ), ( ) max C ( x), C ( y) C x y C x C y Where: C(x, y) represents the size of compressed file obtained by the concatenation of x and y. The NCD can be calculated explicitly between two strings or two files x and y, and this represents how different are these files, facilitating the use of this result into various applications with different data into a parameter-free approach [2, 5] and classify them by unsupervised methods [6]. The NCD is a positive result 0 NCD 1+e, with e as a representation of imperfections of the compression algorithms. It is necessary to remark that the K(x) approximation with C(x) depends on the data with which to work, knowing that common compressors are built base don different hypotheses, some are more efficient than others with a specific type of data. In this article we using a JPEG-LS compressor because the JPEG is a compressor oriented to images, then its performance is higher compared to a general purpose compressor [7]. 3.1 Artifacts Detection Using Rate Distortion Analysis 3. EXPERIMENTS AND RESULTS For the artifacts detection, we propose to use the RD function obtained by compression of the image with different compression factors and examine how an artifact can have a high degree of regularity or irregularity for compression. The RD Analysis is done as the blocks diagram shown in Fig. 6, first we take the image under test, we compress the image with different compression factors, then decompress the image and calculate the error for each compression factor, based on the errors we make a features vector and then apply classification methods. Proc. of SPIE Vol
6 Fig. 6 Block Diagram for Rate-Distortion Analysis: we take the image under test, we compress the image with different compression factors, then decompress the image and calculate the error for each compression factor, based on the errors we make a features vector and then apply classification methods A) Case Study 1 Validation with Synthetic Data:For doing this experiment, we use artificial artifacts in images, so we have simulated the aliasing artifact. For simulating aliasing in an image, is necessary a downsampling process of the image, while we do the downsampling, aliasing occurs and to avoid aliasing, is necessary a low pass filter before downsampling process; taking these considerations, the simulation of aliasing used for this experiment was do following scheme of Fig. 7. Fig. 7 Aliasing Process Simulation, first we make a downsampling process of the image, while we do the downsampling, aliasing occurs and to avoid aliasing, is necessary a low pass filter before downsampling process After applying the methodology presented to the image with aliasing, the results for the aliasing detection is shown in Fig. 8, where we can clearly see the areas that contain the aliasing. (a) Image with Aliasing (b) Aliasing Detection Fig. 8: (a) Shows a satellite image with artificial aliasing. (b) Shows the aliasing detection using RD analysis. Proc. of SPIE Vol
7 B) Case Study 2 Validation with Actual Data:Another result of artifacts with RD, it is the Dropout detection shown in Fig. 9, In this case it is a SPOT image containing actual artifacts, the detection is done correctly. (a) CNES (b) Fig. 9 Dropout (SPOT). (a) Some electronic losses during the image formation process create these randomly saturated pixels. The dropouts often follow a line pattern (corresponding to the structure of the SPOT sensor). (b) Artifact is detected. 3.2 Artifacts Detection Using Normalized Compression Distance (NCD) For this methodology, we take the satellite image and divide it into patches of 64x64 pixels, with these patches we calculate the distance matrix between them using NCD and finally we applied a hierarchical classification method to cluster and identify the patches with artifacts. This process can be seen in Fig. 10. Fig. 10 Experiment Process: we take the satellite image and divide it into patches of 64x64 pixels, with these patches we calculate the distance matrix between them using NCD and finally we applied a hierarchical classification method to cluster and identify the patches with artifacts In this part we will use two datasets, a dataset of satellite images with artificial artifacts introduced manually; and another dataset of satellite images with real artifacts introduced by the sensor itself. A) Case Study 1 Validation with Synthetic Data: For the images with artificial artifacts, we introduce two types of artifacts with different intensities in such a way to study their behavior; these artifacts introduced manually can be seen in Fig. 11, in the first instance we introduce strips with different levels of intensity, which is done is to increase the grayscale value of the satellite image in the desired positions. For the aliasing simulation, we also use different values of down sampling. The introduction of artificial artifacts is made with different intensities to assess the sensitivity of the detection method. Proc. of SPIE Vol
8 Fig. 11 Artifacts Simulation with Various Intensities: the first artifacts are the strips with different levels of intensity, which is done is to increase the grayscale value of the satellite image in the desired positions. The second artifact is aliasing simulation; we also use different values of down sampling. Then, we have a database with different types of artifacts with different intensities and in different environments such as: city, forest and sea; as shown in the following images: Fig. 12 Satellite image database for our experiment with different types of artifacts such as strips and aliasing, with different intensities and in different environments such as city, forest and sea. After making de division of the image into patches of 64x64 pixels as explained in the procedure as shown in Fig. 13, we apply a hierarchical classification method whose results are shown in Fig. 14. The Dendrogram is a type of graphical representation of data as a tree that organizes the data into subcategories that are dividing in others to reach the level of detail desired, this type of representation allows appreciating clearly the relationship between data classes. To plot the dendrogram we use the Euclidean Distance method. Proc. of SPIE Vol
9 Fig. 13 Separate image in patches of 64x64 pixels for calculate the distance matrix between the patches To evaluate the proposed method, we made a study of detection sensitivity with different levels of intensity of the artifact, so also this study has been done in different environments such as: sea, forest and city. So our database contains images of different environments with different types of artifacts at different levels of intensity. Fig. 14 Hierarchical Classification of the Patches: we can see two cluster, red cluster for patches with artifacts, and cyan cluster for patches without artifacts. Proc. of SPIE Vol
10 Finally, after applying the method to our database, the results expressed as percentage of success are: In the table above, we can see the results for the different environment: sea, forest and city, with the strips and the aliasing artifacts in different intensities. The NCD calculation was made using different compressors as JPEG and zip. The intensity of the strips for detection is lower in the sea. The better detection of aliasing occurs in the city because the bandwidth is wider. B) Case Study 2 Validation with Actual Data: For images with real artifacts, we take images acquired by the ROSIS sensor provided by the German Aerospace Center (DLR); these data are hyperspectral images of 7946x512 pixels, 16 bits and 115 bands. We work with subscene of 512x512 pixels as the example in Fig. 15. First thing, we make a manual analysis to determine the location of artifacts, thus we detect strips in the last two bits as shown in Fig. 16. Fig. 15 Examples of 512x512 subscenes with real artifacts, the image is acquired by ROSIS sensor provided by German Aerospace Center (DLR). Proc. of SPIE Vol
11 Fig. 16 Bit by Bit Analysis for Strips Detection in the ROSIS Image: we can see the strips in the last bits. After applying the proposed method, the results are not encouraging and are shown in the following table: These bad results may be because the strips are presented in the last bits are indexed not detected. Given these results, the next experiment is to take directly the binary image containing strips, divided it into 64x64 patches as shown in Fig. 17 and each patches convert to a text string with values of 0s and 1s, then calculate the Normalized Compression Distance with these strings and finally apply the hierarchical classification. For make the conversion to text string, we have two possibilities, a horizontal scanning or vertical scanning as shown in Fig. 18. As each patch has 64x64 pixels and to form the text string we order either row or column after another, then finally each text string will be formed by 4096 values. Fig. 17 Separate binary image in patches of 64x64 pixels for calculate the distance matrix between patches Proc. of SPIE Vol
12 Fig. 18 Formation of the text string based on binary images: we have two options for scanning process, horizontal scanning and vertical scanning. We form the text string order either row or column after another, then finally each text string will be formed by 4096 values The results after to apply the proposed approach is shown in the following table: Horizontal Scanning Vertical Scanning NCD for text Total mixture 81.25% The results for vertical scanning are much better because the strips have the same vertical orientation. Finally we have developed a small tool to do an artifact detection process using the previously methods reported, the interface tool is shown in Fig. 19. This interface has the methods proposed, both, the calculation of NCD as also the use of Rate-Distortion function; in this interface it is possible to vary the size of patches. Fig. 19 Tool for Artifacts Detection: With this interface we can apply the different methods proposed, both, the calculation of NCD as also use the Rate-Distortion analysis; in this interface it is possible to vary the size of patches Proc. of SPIE Vol
13 4. CONCLUSIONS The conclusions are in function of two datasets: About the artificial artifacts We can appreciate that the strips have an acceptable possibility to be detected in a sea environment from an intensity of k = 10, however in the forest and the city environment with intensity k = 30. The aliasing can be detected in a city environment, but not at sea or in the forest, due to de city s bandwidth is widest than the sea and forest. The detection of artifacts is done best way depending on the environment we are working, at sea and in the field is easier to detect the strips but not so the aliasing, while in the city and the forest is easier to detect aliasing. About the real artifacts We can see that there aren t good results and this may be because the strips are presented in the last bits and can not be detected. The acceptable results are found with scanning vertically, column by column as the strips also have the same orientation. 5. ACKNOWLEDGMENT HyMap was made available from HyVista Corp. and DLRs Optical Airborne Remote Sensing and Calibration Facility service ( and the ROSIS data by DLRs OpAIRS service ( The authors very much acknowledge the support of Dr. Martin Bachmann from DLR. REFERENCES [1] J. Hyung-Sup, W. Joong-Sun, K. Myung-Ho, L. Yong-Woong, Detection and Restoration of Defective Lines in the SPOT 4 SWIR Band, IEEE Transaction on Image Processing, [2] D. Cerra, A. Mallet, L. Gueguen, M. Datcu, Algorithmic Information Theory Based Analysis of Earth Observation Images: an Assessment, IEEE Geosciences and remote Sensing Letters, in press. [3] A. Mallet, M. Datcu, Rate Distortion Based Detection of Artifacts in Earth Observation Images, IEEE Geosciences and Remote Sensing Letters, vol. 5, N 3, pp , July [4] M. Li, X. Chen, X. Li, B. Ma, P. M. B. Vitanyi; The Similarity Metric, IEEE Trans. Inf. Theory, vol. 50, N 12, pp [5] E. Keogh, S. Lonardi, Ch. Ratanamahatana, Towards Parameter-Free Data Mining, Department of Computer Science and Engineering, University of California, Riverside. [6] R. Cilibrasi, P. M. B. Vitanyi; Clustering by Compression, IEEE Transaction on Information Theory, vol. 51, N 4, April 2005, pp [7] T. Tao, A. Mukherjee, and R.V. Satya, A search-aware JPEG-LS variation for compressed image retrieval, Intelligent Multimedia, Video and Speech Processing, (2004), Proc. of SPIE Vol
Image Classification Using Data Compression Techniques
Image Classification Using Data Compression Techniques Martha Roxana Quispe Ayala, Krista Asalde Alvarez, Avid Roman Gonzalez To cite this version: Martha Roxana Quispe Ayala, Krista Asalde Alvarez, Avid
More informationModelling image complexity by independent component analysis, with application to content-based image retrieval
Modelling image complexity by independent component analysis, with application to content-based image retrieval Jukka Perkiö 1,2 and Aapo Hyvärinen 1,2,3 1 Helsinki Institute for Information Technology,
More informationThe SNCD as a Metrics for Image Quality Assessment
The SNCD as a Metrics for Image Qualit Assessment Avid Roman-Gonzalez TELECOM ParisTech, Departement TSI Paris, France Abstract In our era, when we have a lot of instrument to capture digital images and
More informationADAPTIVE WINDOWING FOR OPTIMAL VISUALIZATION OF MEDICAL IMAGES BASED ON NORMALIZED INFORMATION DISTANCE
ADAPTIVE WINDOWING FOR OPTIMAL VISUALIZATION OF MEDICAL IMAGES BASED ON NORMALIZED INFORMATION DISTANCE Nima Nikvand, Hojatollah Yeganeh and Zhou Wang Dept. of Electrical & Computer Engineering, University
More informationA Image Comparative Study using DCT, Fast Fourier, Wavelet Transforms and Huffman Algorithm
International Journal of Engineering Research and General Science Volume 3, Issue 4, July-August, 15 ISSN 91-2730 A Image Comparative Study using DCT, Fast Fourier, Wavelet Transforms and Huffman Algorithm
More informationScanner Parameter Estimation Using Bilevel Scans of Star Charts
ICDAR, Seattle WA September Scanner Parameter Estimation Using Bilevel Scans of Star Charts Elisa H. Barney Smith Electrical and Computer Engineering Department Boise State University, Boise, Idaho 8375
More informationFast Sample Generation with Variational Bayesian for Limited Data Hyperspectral Image Classification
Fast Sample Generation with Variational Bayesian for Limited Data Hyperspectral Image Classification July 26, 2018 AmirAbbas Davari, Hasan Can Özkan, Andreas Maier, Christian Riess Pattern Recognition
More informationCompression-Based Clustering of Chromagram Data: New Method and Representations
Compression-Based Clustering of Chromagram Data: New Method and Representations Teppo E. Ahonen Department of Computer Science University of Helsinki teahonen@cs.helsinki.fi Abstract. We approach the problem
More informationDesign and Use of. Earth Observation Image Content Tools
Design and Use of Earth Observation Image Content Tools Mihai Datcu (1, 2), Daniele Cerra (1), Houda Chaabouni-Chouayakh (1), Amaia de Miguel (1), Daniela Espinoza Molina (1), Gottfried Schwarz (1), Matteo
More informationNearest Clustering Algorithm for Satellite Image Classification in Remote Sensing Applications
Nearest Clustering Algorithm for Satellite Image Classification in Remote Sensing Applications Anil K Goswami 1, Swati Sharma 2, Praveen Kumar 3 1 DRDO, New Delhi, India 2 PDM College of Engineering for
More informationSTEREO EVALUATION OF ALOS/PRISM DATA ON ESA-AO TEST SITES FIRST DLR RESULTS
STEREO EVALUATION OF ALOS/PRISM DATA ON ESA-AO TEST SITES FIRST DLR RESULTS Authors: Mathias Schneider, Manfred Lehner, Rupert Müller, Peter Reinartz Remote Sensing Technology Institute German Aerospace
More informationCompression of RADARSAT Data with Block Adaptive Wavelets Abstract: 1. Introduction
Compression of RADARSAT Data with Block Adaptive Wavelets Ian Cumming and Jing Wang Department of Electrical and Computer Engineering The University of British Columbia 2356 Main Mall, Vancouver, BC, Canada
More informationAn ICA based Approach for Complex Color Scene Text Binarization
An ICA based Approach for Complex Color Scene Text Binarization Siddharth Kherada IIIT-Hyderabad, India siddharth.kherada@research.iiit.ac.in Anoop M. Namboodiri IIIT-Hyderabad, India anoop@iiit.ac.in
More informationSession 2 A virtual Observatory for TerraSAR-X data
Session 2 A virtual Observatory for TerraSAR-X data 2nd User Community Workshop Darmstadt, 10-11 May 2012 Presenter: Mihai Datcu and Daniela Espinoza Molina (DLR) This presentation contains contributions
More informationThe Gain setting for Landsat 7 (High or Low Gain) depends on: Sensor Calibration - Application. the surface cover types of the earth and the sun angle
Sensor Calibration - Application Station Identifier ASN Scene Center atitude 34.840 (34 3'0.64"N) Day Night DAY Scene Center ongitude 33.03270 (33 0'7.72"E) WRS Path WRS Row 76 036 Corner Upper eft atitude
More informationScanner Parameter Estimation Using Bilevel Scans of Star Charts
Boise State University ScholarWorks Electrical and Computer Engineering Faculty Publications and Presentations Department of Electrical and Computer Engineering 1-1-2001 Scanner Parameter Estimation Using
More information(Refer Slide Time: 0:51)
Introduction to Remote Sensing Dr. Arun K Saraf Department of Earth Sciences Indian Institute of Technology Roorkee Lecture 16 Image Classification Techniques Hello everyone welcome to 16th lecture in
More informationSession 2 A virtual Observatory for TerraSAR-X data
Session 2 A virtual Observatory for TerraSAR-X data 3rd User Community Workshop Chania, 12-14 June 2013 Presenter: Mihai Datcu and Daniela Espinoza Molina (DLR) Motivation ENVISAT provided measurements
More informationPRINCIPAL COMPONENT ANALYSIS IMAGE DENOISING USING LOCAL PIXEL GROUPING
PRINCIPAL COMPONENT ANALYSIS IMAGE DENOISING USING LOCAL PIXEL GROUPING Divesh Kumar 1 and Dheeraj Kalra 2 1 Department of Electronics & Communication Engineering, IET, GLA University, Mathura 2 Department
More informationRegion Based Image Fusion Using SVM
Region Based Image Fusion Using SVM Yang Liu, Jian Cheng, Hanqing Lu National Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Sciences ABSTRACT This paper presents a novel
More informationRobust Steganography Using Texture Synthesis
Robust Steganography Using Texture Synthesis Zhenxing Qian 1, Hang Zhou 2, Weiming Zhang 2, Xinpeng Zhang 1 1. School of Communication and Information Engineering, Shanghai University, Shanghai, 200444,
More informationEVALUATION OF CONVENTIONAL DIGITAL CAMERA SCENES FOR THEMATIC INFORMATION EXTRACTION ABSTRACT
EVALUATION OF CONVENTIONAL DIGITAL CAMERA SCENES FOR THEMATIC INFORMATION EXTRACTION H. S. Lim, M. Z. MatJafri and K. Abdullah School of Physics Universiti Sains Malaysia, 11800 Penang ABSTRACT A study
More informationReconstructing Images of Bar Codes for Construction Site Object Recognition 1
Reconstructing Images of Bar Codes for Construction Site Object Recognition 1 by David E. Gilsinn 2, Geraldine S. Cheok 3, Dianne P. O Leary 4 ABSTRACT: This paper discusses a general approach to reconstructing
More informationImage Inpainting by Hyperbolic Selection of Pixels for Two Dimensional Bicubic Interpolations
Image Inpainting by Hyperbolic Selection of Pixels for Two Dimensional Bicubic Interpolations Mehran Motmaen motmaen73@gmail.com Majid Mohrekesh mmohrekesh@yahoo.com Mojtaba Akbari mojtaba.akbari@ec.iut.ac.ir
More informationImage Quality Assessment Techniques: An Overview
Image Quality Assessment Techniques: An Overview Shruti Sonawane A. M. Deshpande Department of E&TC Department of E&TC TSSM s BSCOER, Pune, TSSM s BSCOER, Pune, Pune University, Maharashtra, India Pune
More informationSpatially variant dimensionality reduction for the visualization of multi/hyperspectral images
Author manuscript, published in "International Conference on Image Analysis and Recognition, Burnaby : Canada (2011)" DOI : 10.1007/978-3-642-21593-3_38 Spatially variant dimensionality reduction for the
More informationADVANCED IMAGE PROCESSING METHODS FOR ULTRASONIC NDE RESEARCH C. H. Chen, University of Massachusetts Dartmouth, N.
ADVANCED IMAGE PROCESSING METHODS FOR ULTRASONIC NDE RESEARCH C. H. Chen, University of Massachusetts Dartmouth, N. Dartmouth, MA USA Abstract: The significant progress in ultrasonic NDE systems has now
More informationElemental Set Methods. David Banks Duke University
Elemental Set Methods David Banks Duke University 1 1. Introduction Data mining deals with complex, high-dimensional data. This means that datasets often combine different kinds of structure. For example:
More informationImage Compression Algorithm and JPEG Standard
International Journal of Scientific and Research Publications, Volume 7, Issue 12, December 2017 150 Image Compression Algorithm and JPEG Standard Suman Kunwar sumn2u@gmail.com Summary. The interest in
More informationIntroduction to digital image classification
Introduction to digital image classification Dr. Norman Kerle, Wan Bakx MSc a.o. INTERNATIONAL INSTITUTE FOR GEO-INFORMATION SCIENCE AND EARTH OBSERVATION Purpose of lecture Main lecture topics Review
More informationSemi-Automatic Transcription Tool for Ancient Manuscripts
The Venice Atlas A Digital Humanities atlas project by DH101 EPFL Students Semi-Automatic Transcription Tool for Ancient Manuscripts In this article, we investigate various techniques from the fields of
More informationUnsupervised Change Detection in Optical Satellite Images using Binary Descriptor
Unsupervised Change Detection in Optical Satellite Images using Binary Descriptor Neha Gupta, Gargi V. Pillai, Samit Ari Department of Electronics and Communication Engineering, National Institute of Technology,
More informationK-Means Clustering 3/3/17
K-Means Clustering 3/3/17 Unsupervised Learning We have a collection of unlabeled data points. We want to find underlying structure in the data. Examples: Identify groups of similar data points. Clustering
More informationChange Detection in Remotely Sensed Images Based on Image Fusion and Fuzzy Clustering
International Journal of Electronics Engineering Research. ISSN 0975-6450 Volume 9, Number 1 (2017) pp. 141-150 Research India Publications http://www.ripublication.com Change Detection in Remotely Sensed
More informationAUTOMATIC INTERPRETATION OF HIGH RESOLUTION SAR IMAGES: FIRST RESULTS OF SAR IMAGE SIMULATION FOR SINGLE BUILDINGS
AUTOMATIC INTERPRETATION OF HIGH RESOLUTION SAR IMAGES: FIRST RESULTS OF SAR IMAGE SIMULATION FOR SINGLE BUILDINGS J. Tao *, G. Palubinskas, P. Reinartz German Aerospace Center DLR, 82234 Oberpfaffenhofen,
More informationTexture Segmentation by Windowed Projection
Texture Segmentation by Windowed Projection 1, 2 Fan-Chen Tseng, 2 Ching-Chi Hsu, 2 Chiou-Shann Fuh 1 Department of Electronic Engineering National I-Lan Institute of Technology e-mail : fctseng@ccmail.ilantech.edu.tw
More informationA Method for the Identification of Inaccuracies in Pupil Segmentation
A Method for the Identification of Inaccuracies in Pupil Segmentation Hugo Proença and Luís A. Alexandre Dep. Informatics, IT - Networks and Multimedia Group Universidade da Beira Interior, Covilhã, Portugal
More informationMobile Human Detection Systems based on Sliding Windows Approach-A Review
Mobile Human Detection Systems based on Sliding Windows Approach-A Review Seminar: Mobile Human detection systems Njieutcheu Tassi cedrique Rovile Department of Computer Engineering University of Heidelberg
More informationGABOR AND WEBER FEATURE EXTRACTION PERFORMANCE BASED ON URBAN ATLAS GROUND TRUTH
U.P.B. Sci. Bull., Series C, Vol. 78, Iss. 3, 2016 ISSN 2286-3540 GABOR AND WEBER FEATURE EXTRACTION PERFORMANCE BASED ON URBAN ATLAS GROUND TRUTH Mihaela STAN 1, Anca POPESCU 2, Dan Alexandru STOICHESCU
More informationFace Detection using Hierarchical SVM
Face Detection using Hierarchical SVM ECE 795 Pattern Recognition Christos Kyrkou Fall Semester 2010 1. Introduction Face detection in video is the process of detecting and classifying small images extracted
More informationCompressive Sensing for Multimedia. Communications in Wireless Sensor Networks
Compressive Sensing for Multimedia 1 Communications in Wireless Sensor Networks Wael Barakat & Rabih Saliba MDDSP Project Final Report Prof. Brian L. Evans May 9, 2008 Abstract Compressive Sensing is an
More informationA Framework of Hyperspectral Image Compression using Neural Networks
A Framework of Hyperspectral Image Compression using Neural Networks Yahya M. Masalmah, Ph.D 1, Christian Martínez-Nieves 1, Rafael Rivera-Soto 1, Carlos Velez 1, and Jenipher Gonzalez 1 1 Universidad
More informationScalable Coding of Image Collections with Embedded Descriptors
Scalable Coding of Image Collections with Embedded Descriptors N. Adami, A. Boschetti, R. Leonardi, P. Migliorati Department of Electronic for Automation, University of Brescia Via Branze, 38, Brescia,
More informationComparative Study of Hand Gesture Recognition Techniques
Reg. No.:20140316 DOI:V2I4P16 Comparative Study of Hand Gesture Recognition Techniques Ann Abraham Babu Information Technology Department University of Mumbai Pillai Institute of Information Technology
More informationDepartment of Electronics and Communication KMP College of Engineering, Perumbavoor, Kerala, India 1 2
Vol.3, Issue 3, 2015, Page.1115-1021 Effect of Anti-Forensics and Dic.TV Method for Reducing Artifact in JPEG Decompression 1 Deepthy Mohan, 2 Sreejith.H 1 PG Scholar, 2 Assistant Professor Department
More informationHyperspectral Image Segmentation using Homogeneous Area Limiting and Shortest Path Algorithm
Hyperspectral Image Segmentation using Homogeneous Area Limiting and Shortest Path Algorithm Fatemeh Hajiani Department of Electrical Engineering, College of Engineering, Khormuj Branch, Islamic Azad University,
More informationDirection-Length Code (DLC) To Represent Binary Objects
IOSR Journal of Computer Engineering (IOSR-JCE) e-issn: 2278-0661,p-ISSN: 2278-8727, Volume 18, Issue 2, Ver. I (Mar-Apr. 2016), PP 29-35 www.iosrjournals.org Direction-Length Code (DLC) To Represent Binary
More informationANALYSIS OF SPIHT ALGORITHM FOR SATELLITE IMAGE COMPRESSION
ANALYSIS OF SPIHT ALGORITHM FOR SATELLITE IMAGE COMPRESSION K Nagamani (1) and AG Ananth (2) (1) Assistant Professor, R V College of Engineering, Bangalore-560059. knmsm_03@yahoo.com (2) Professor, R V
More informationKeywords Clustering, Goals of clustering, clustering techniques, clustering algorithms.
Volume 3, Issue 5, May 2013 ISSN: 2277 128X International Journal of Advanced Research in Computer Science and Software Engineering Research Paper Available online at: www.ijarcsse.com A Survey of Clustering
More informationEnhanced Image. Improved Dam point Labelling
3rd International Conference on Multimedia Technology(ICMT 2013) Video Text Extraction Based on Stroke Width and Color Xiaodong Huang, 1 Qin Wang, Kehua Liu, Lishang Zhu Abstract. Video text can be used
More informationUNIT-2 IMAGE REPRESENTATION IMAGE REPRESENTATION IMAGE SENSORS IMAGE SENSORS- FLEX CIRCUIT ASSEMBLY
18-08-2016 UNIT-2 In the following slides we will consider what is involved in capturing a digital image of a real-world scene Image sensing and representation Image Acquisition Sampling and quantisation
More informationPCA vs. ICA Decomposition of High Resolution SAR Images: Application to Urban Structures Recognition
PCA vs. ICA Decomposition of High Resolution SAR Images: Application to Urban Structures Recognition Houda Chaabouni-Chouayakh a and Mihai Datcu a,b a German Aerospace Center (DLR), Oberpfaffenhofen D-82234
More informationContextual High-Resolution Image Classification by Markovian Data Fusion, Adaptive Texture Extraction, and Multiscale Segmentation
IGARSS-2011 Vancouver, Canada, July 24-29, 29, 2011 Contextual High-Resolution Image Classification by Markovian Data Fusion, Adaptive Texture Extraction, and Multiscale Segmentation Gabriele Moser Sebastiano
More informationEnhancing DubaiSat-1 Satellite Imagery Using a Single Image Super-Resolution
Enhancing DubaiSat-1 Satellite Imagery Using a Single Image Super-Resolution Saeed AL-Mansoori 1 and Alavi Kunhu 2 1 Associate Image Processing Engineer, SIPAD Image Enhancement Section Emirates Institution
More informationCHAPTER 4 DETECTION OF DISEASES IN PLANT LEAF USING IMAGE SEGMENTATION
CHAPTER 4 DETECTION OF DISEASES IN PLANT LEAF USING IMAGE SEGMENTATION 4.1. Introduction Indian economy is highly dependent of agricultural productivity. Therefore, in field of agriculture, detection of
More informationPATTERN CLASSIFICATION AND SCENE ANALYSIS
PATTERN CLASSIFICATION AND SCENE ANALYSIS RICHARD O. DUDA PETER E. HART Stanford Research Institute, Menlo Park, California A WILEY-INTERSCIENCE PUBLICATION JOHN WILEY & SONS New York Chichester Brisbane
More informationHierarchical Minimum Spanning Trees for Lossy Image Set Compression
Hierarchical Minimum Spanning Trees for Lossy Image Set Compression Anthony Schmieder, Barry Gergel, and Howard Cheng Department of Mathematics and Computer Science University of Lethbridge, Alberta, Canada
More informationVisible and Long-Wave Infrared Image Fusion Schemes for Situational. Awareness
Visible and Long-Wave Infrared Image Fusion Schemes for Situational Awareness Multi-Dimensional Digital Signal Processing Literature Survey Nathaniel Walker The University of Texas at Austin nathaniel.walker@baesystems.com
More informationScene Text Detection Using Machine Learning Classifiers
601 Scene Text Detection Using Machine Learning Classifiers Nafla C.N. 1, Sneha K. 2, Divya K.P. 3 1 (Department of CSE, RCET, Akkikkvu, Thrissur) 2 (Department of CSE, RCET, Akkikkvu, Thrissur) 3 (Department
More informationArtificial Neural Networks Lab 2 Classical Pattern Recognition
Artificial Neural Networks Lab 2 Classical Pattern Recognition Purpose To implement (using MATLAB) some traditional classifiers and try these on both synthetic and real data. The exercise should give some
More informationRobust color segmentation algorithms in illumination variation conditions
286 CHINESE OPTICS LETTERS / Vol. 8, No. / March 10, 2010 Robust color segmentation algorithms in illumination variation conditions Jinhui Lan ( ) and Kai Shen ( Department of Measurement and Control Technologies,
More informationKeywords Data compression, image processing, NCD, Kolmogorov complexity, JPEG, ZIP
Volume 3, Issue 7, July 203 ISSN: 2277 28X International Journal of Avance Research in Computer Science an Software Engineering Research Paper Available online at: wwwijarcssecom Compression Techniques
More informationKES: Knowledge Enabled Services for better EO Information Use. Andrea Colapicchioni Advanced Computer Systems Space Division
KES: Knowledge Enabled Services for better EO Information Use Andrea Colapicchioni Advanced Computer Systems Space Division a.colapicchioni@acsys.it The problem During the last decades, the satellite image
More information(Refer Slide Time 00:17) Welcome to the course on Digital Image Processing. (Refer Slide Time 00:22)
Digital Image Processing Prof. P. K. Biswas Department of Electronics and Electrical Communications Engineering Indian Institute of Technology, Kharagpur Module Number 01 Lecture Number 02 Application
More informationINCREASING CLASSIFICATION QUALITY BY USING FUZZY LOGIC
JOURNAL OF APPLIED ENGINEERING SCIENCES VOL. 1(14), issue 4_2011 ISSN 2247-3769 ISSN-L 2247-3769 (Print) / e-issn:2284-7197 INCREASING CLASSIFICATION QUALITY BY USING FUZZY LOGIC DROJ Gabriela, University
More informationData Hiding in Video
Data Hiding in Video J. J. Chae and B. S. Manjunath Department of Electrical and Computer Engineering University of California, Santa Barbara, CA 9316-956 Email: chaejj, manj@iplab.ece.ucsb.edu Abstract
More informationThree-Dimensional Computer Vision
\bshiaki Shirai Three-Dimensional Computer Vision With 313 Figures ' Springer-Verlag Berlin Heidelberg New York London Paris Tokyo Table of Contents 1 Introduction 1 1.1 Three-Dimensional Computer Vision
More informationSummer Internship at STMicroelectronics, Greater Noida, India
Summer Internship at STMicroelectronics, Greater Noida, India Thibault Lefeuvre, Ecole Polytechnique, France August 2015 Techniques of frame compression in ultra low power encoder Introduction At first,
More informationUAV-based Remote Sensing Payload Comprehensive Validation System
36th CEOS Working Group on Calibration and Validation Plenary May 13-17, 2013 at Shanghai, China UAV-based Remote Sensing Payload Comprehensive Validation System Chuan-rong LI Project PI www.aoe.cas.cn
More informationMODULE 3 LECTURE NOTES 3 ATMOSPHERIC CORRECTIONS
MODULE 3 LECTURE NOTES 3 ATMOSPHERIC CORRECTIONS 1. Introduction The energy registered by the sensor will not be exactly equal to that emitted or reflected from the terrain surface due to radiometric and
More informationNew Edge-Enhanced Error Diffusion Algorithm Based on the Error Sum Criterion
New Edge-Enhanced Error Diffusion Algorithm Based on the Error Sum Criterion Jae Ho Kim* Tae Il Chung Hyung Soon Kim* Kyung Sik Son* Pusan National University Image and Communication Laboratory San 3,
More informationDefect Detection of Regular Patterned Fabric by Spectral Estimation Technique and Rough Set Classifier
Defect Detection of Regular Patterned Fabric by Spectral Estimation Technique and Rough Set Classifier Mr..Sudarshan Deshmukh. Department of E&TC Siddhant College of Engg, Sudumbare, Pune Prof. S. S. Raut.
More informationHighly Secure Invertible Data Embedding Scheme Using Histogram Shifting Method
www.ijecs.in International Journal Of Engineering And Computer Science ISSN:2319-7242 Volume 3 Issue 8 August, 2014 Page No. 7932-7937 Highly Secure Invertible Data Embedding Scheme Using Histogram Shifting
More informationUsing Structural Features to Detect Buildings in Panchromatic Satellite Images
Using Structural Features to Detect Buildings in Panchromatic Satellite Images Beril Sırmaçek German Aerospace Center (DLR) Remote Sensing Technology Institute Weßling, 82234, Germany E-mail: Beril.Sirmacek@dlr.de
More informationImage Processing and Analysis
Image Processing and Analysis 3 stages: Image Restoration - correcting errors and distortion. Warping and correcting systematic distortion related to viewing geometry Correcting "drop outs", striping and
More informationUnsupervised and Self-taught Learning for Remote Sensing Image Analysis
Unsupervised and Self-taught Learning for Remote Sensing Image Analysis Ribana Roscher Institute of Geodesy and Geoinformation, Remote Sensing Group, University of Bonn 1 The Changing Earth https://earthengine.google.com/timelapse/
More informationRemote Sensing Image Analysis via a Texture Classification Neural Network
Remote Sensing Image Analysis via a Texture Classification Neural Network Hayit K. Greenspan and Rodney Goodman Department of Electrical Engineering California Institute of Technology, 116-81 Pasadena,
More informationAnalysis of Dendrogram Tree for Identifying and Visualizing Trends in Multi-attribute Transactional Data
Analysis of Dendrogram Tree for Identifying and Visualizing Trends in Multi-attribute Transactional Data D.Radha Rani 1, A.Vini Bharati 2, P.Lakshmi Durga Madhuri 3, M.Phaneendra Babu 4, A.Sravani 5 Department
More informationAnalisi di immagini iperspettrali satellitari multitemporali: metodi ed applicazioni
Analisi di immagini iperspettrali satellitari multitemporali: metodi ed applicazioni E-mail: bovolo@fbk.eu Web page: http://rsde.fbk.eu Outline 1 Multitemporal image analysis 2 Multitemporal images pre-processing
More informationLecture 10: Semantic Segmentation and Clustering
Lecture 10: Semantic Segmentation and Clustering Vineet Kosaraju, Davy Ragland, Adrien Truong, Effie Nehoran, Maneekwan Toyungyernsub Department of Computer Science Stanford University Stanford, CA 94305
More informationA Miniature-Based Image Retrieval System
A Miniature-Based Image Retrieval System Md. Saiful Islam 1 and Md. Haider Ali 2 Institute of Information Technology 1, Dept. of Computer Science and Engineering 2, University of Dhaka 1, 2, Dhaka-1000,
More informationHybrid Video Compression Using Selective Keyframe Identification and Patch-Based Super-Resolution
2011 IEEE International Symposium on Multimedia Hybrid Video Compression Using Selective Keyframe Identification and Patch-Based Super-Resolution Jeffrey Glaister, Calvin Chan, Michael Frankovich, Adrian
More informationAdaptive Normal Map Compression for 3D Video Games
Adaptive Normal Map Compression for 3D Video Games Alexander Wong a28wong@engmail.uwaterloo.ca William Bishop wdbishop@uwaterloo.ca Department of Electrical and Computer Engineering University of Waterloo
More informationVisualization and text mining of patent and non-patent data
of patent and non-patent data Anton Heijs Information Solutions Delft, The Netherlands http://www.treparel.com/ ICIC conference, Nice, France, 2008 Outline Introduction Applications on patent and non-patent
More informationAn Efficient Approach for Color Pattern Matching Using Image Mining
An Efficient Approach for Color Pattern Matching Using Image Mining * Manjot Kaur Navjot Kaur Master of Technology in Computer Science & Engineering, Sri Guru Granth Sahib World University, Fatehgarh Sahib,
More informationStructured Light. Tobias Nöll Thanks to Marc Pollefeys, David Nister and David Lowe
Structured Light Tobias Nöll tobias.noell@dfki.de Thanks to Marc Pollefeys, David Nister and David Lowe Introduction Previous lecture: Dense reconstruction Dense matching of non-feature pixels Patch-based
More informationMeasuring Ground Deformation using Optical Imagery
Measuring Ground Deformation using Optical Imagery Sébastien Leprince California Institute of Technology, USA October 29, 2009 Keck Institute for Space Studies Workshop Measuring Horizontal Ground Displacement,
More informationSpatial Information Based Image Classification Using Support Vector Machine
Spatial Information Based Image Classification Using Support Vector Machine P.Jeevitha, Dr. P. Ganesh Kumar PG Scholar, Dept of IT, Regional Centre of Anna University, Coimbatore, India. Assistant Professor,
More informationPeriodicity Extraction using Superposition of Distance Matching Function and One-dimensional Haar Wavelet Transform
Periodicity Extraction using Superposition of Distance Matching Function and One-dimensional Haar Wavelet Transform Dr. N.U. Bhajantri Department of Computer Science & Engineering, Government Engineering
More informationContent Based Image Retrieval: Survey and Comparison between RGB and HSV model
Content Based Image Retrieval: Survey and Comparison between RGB and HSV model Simardeep Kaur 1 and Dr. Vijay Kumar Banga 2 AMRITSAR COLLEGE OF ENGG & TECHNOLOGY, Amritsar, India Abstract Content based
More informationKapitel 4: Clustering
Ludwig-Maximilians-Universität München Institut für Informatik Lehr- und Forschungseinheit für Datenbanksysteme Knowledge Discovery in Databases WiSe 2017/18 Kapitel 4: Clustering Vorlesung: Prof. Dr.
More informationFUZZY FOREST LEARNING BASED ONLINE FACIAL BIOMETRIC VERIFICATION FOR PRIVACY PROTECTION
FUZZY FOREST LEARNING BASED ONLINE FACIAL BIOMETRIC VERIFICATION FOR PRIVACY PROTECTION Supritha G M 1, Shivanand R D 2 1 P.G. Student, Dept. of Computer Science and Engineering, B.I.E.T College, Karnataka,
More informationPower Functions and Their Use In Selecting Distance Functions for. Document Degradation Model Validation. 600 Mountain Avenue, Room 2C-322
Power Functions and Their Use In Selecting Distance Functions for Document Degradation Model Validation Tapas Kanungo y ; Robert M. Haralick y and Henry S. Baird z y Department of Electrical Engineering,
More informationOptimizing the Deblocking Algorithm for. H.264 Decoder Implementation
Optimizing the Deblocking Algorithm for H.264 Decoder Implementation Ken Kin-Hung Lam Abstract In the emerging H.264 video coding standard, a deblocking/loop filter is required for improving the visual
More informationImage Similarity Using Sparse Representation and Compression Distance
JOURNAL, VOL. X, NO. X, APR 2013 1 Image Similarity Using Sparse Representation and Compression Distance Tanaya Guha, Student Member, IEEE, and Rabab K Ward, Fellow, IEEE arxiv:1206.2627v2 [cs.cv] 7 May
More informationClustering Analysis based on Data Mining Applications Xuedong Fan
Applied Mechanics and Materials Online: 203-02-3 ISSN: 662-7482, Vols. 303-306, pp 026-029 doi:0.4028/www.scientific.net/amm.303-306.026 203 Trans Tech Publications, Switzerland Clustering Analysis based
More informationHandwritten Script Recognition at Block Level
Chapter 4 Handwritten Script Recognition at Block Level -------------------------------------------------------------------------------------------------------------------------- Optical character recognition
More informationImage Compression with Competitive Networks and Pre-fixed Prototypes*
Image Compression with Competitive Networks and Pre-fixed Prototypes* Enrique Merida-Casermeiro^, Domingo Lopez-Rodriguez^, and Juan M. Ortiz-de-Lazcano-Lobato^ ^ Department of Applied Mathematics, University
More informationLocating the Shadow Regions in LiDAR Data: Results on the SHARE 2012 Dataset
Locating the Shadow Regions in LiDAR Data: Results on the SHARE 22 Dataset Mustafa BOYACI, Seniha Esen YUKSEL* Hacettepe University, Department of Electrical and Electronics Engineering Beytepe, Ankara,
More informationA Laplacian Based Novel Approach to Efficient Text Localization in Grayscale Images
A Laplacian Based Novel Approach to Efficient Text Localization in Grayscale Images Karthik Ram K.V & Mahantesh K Department of Electronics and Communication Engineering, SJB Institute of Technology, Bangalore,
More information